Building a datalab service

Talk during RStudio & thinkR Roadshow

Disclaimer: The views and opinions expressed in this presentation are those of the author and do not necessarly reflect the ones from his employer.

Why a datalab with R can help you promote good pratices in a corporate environment ?
How it can help to be more productive and ease the path to R in production ?
How to build a service on top a good technical fundation with reliable products ?
How it helps create a community inside a corporate environment ?

Transcript

1. Building a datalab service with RStudio products RStudio roadshow with Thinkr Christophe Dervieux 6 June 2019
2. About me The 3 W Who: R user & developer Background in engineering & economics in the energy sector. Where: Currently working at RTE France What: Analytic Admin & Developer Advocate supporting datascience projects and teams. Disclaimer: The views and opinions expressed in this presentation are those of the author and do not necessarly reflect the ones from his employer. 2 / 16
3. Why a datalab ? Working with data in corporate environment Everyone is working with data Activities are more and more data centric Data is more accessible Coporate environment can feel special Specific and secure environments are required Integration with the company ecosystem is key (toward R data products into production) Common solutions and common practices can help improve. 3 / 16
4. Why a datalab with R ? Leveraging a widely used tool R as a common tool Already used by some experts (ex: Datascience and R&D teams) Everyone can benefit (About Excel and its limits) A large community to build on top Instead of learning Advanced Excel use and VBA language, just learn R instead Promote R as a tool for everyone to work with data 4 / 16
5. Why a datalab with R ? Solid ground is available Ready-to-use professional products Professional products by Rstudio to complement the available open source tooling Investment by RStudio into the R for enterprise vision and into the open-source ecosystem Why this is important ? Easier to integrate in a non data-centric company Professional support available (not to underestimate) Time to focus on specific value for the company ecosystem 5 / 16
6. Key features What to aim for Shared and integrated environment Connected to company databases and other services Allow better collaboration between teams Shared ressources for demanding computation Ready-to-use environment No configuration and installation required Available from beginners to advanced users Just connect and do your job 6 / 16
7. What does it look like ? One possible setup 7 / 16
8. Is it enough ? Offering more than a technical platform Building only a technical platform is not enough but it is the required first step. - it must be helpful How to improve user experience in a coporate environment ? How to add value so that users want to work with it ? - it must be used How to attract users and promote R as an alternative ? How to accompagny change toward new practices ? Tooling + skilled people = offering specific services 8 / 16
9. So what is to offer ? with help from a datalab team and Rstudio Products 9 / 16
10. Improve user experience Thanks to RStudio IDE Extension mechanism Custom internal R packages for the datalab Company templates for projects and documents Specific configurations to fit everyone needs (by user, by group, by project) Custom RStudio Addins An average R user should not be a developer or have necessarly IT skills... Could but certainly not supposed to Aim: Add value compared to working on its own local environment 10 / 16
11. Advices & Support Be there for the users Providing additional documentations about how to use some packages inside corporate ecosytem Supporting for any R related subjects Being involved on projects to help with a good start Watching closely the community and being involved A user can't google about your datalab, you need to provide specific answers Aim: User centric approach: Being the first user to better accompany changes! 11 / 16
12. Trainings & Demos Empowering the user On-premise training using the datalab Sharing recipes as html document, learnr tutorial or bookdown e-learning and published inside the datalab Rmakdown website to centralize documentation, contacts and Q&A Showcase, Gallery, Screencast and videos Showing how to take advantage of the datalab services Aim: Share good practices, teach new skills, support changes, build a community 12 / 16
13. So what is to offer ? with help from a datalab team and Rstudio Product Being there for guidance on methods and tooling Letting the users focus on business cases and field expertise 13 / 16
14. So what is the catch ? Advices from my experience Dedicated people to manage the datalab offer An R-admin to bridge "the gap between datascience and IT" Challenging, specifically in a not-so-flexible IT ecosystem Empowering the users Building with reliable products & Focus on the datalab service 14 / 16
15. So what is the catch ? Does it work ? R in enterprise can be a thing R can become a known technology in the IT ecosystem. A datalab allows closer integration with other solutions in the company. Ease the path toward R into productions. 15 / 16
16.  @cderv  @chrisderv Thanks ! Merci ! Questions ?